Trustworthy Verification and Visualisation of Multiple XML

Trustworthy Verification and Visualisation of Multiple XML-Signatures Wolfgang Kubbilun1, Sebastian Gajek2, Michael Psarros2, and Jörg Schwenk2 1 MediaSec Technologies GmbH, Berliner Platz 6-8, 45127 Essen, Germany [email protected] 2 Horst Görtz Institute for IT Security, Ruhr-University Bochum, Germany {sebastian.gajek, michael.psarros, joerg.schwenk}@nds.rub.de Abstract. The digital signature is one of the most important cryptographic primitives. It provides data integrity, message authentication and non-repudiation, which are required attributes in security critical services, such as electronic commerce, voting or health care. Whereas previous data formats for digital signatures concentrated on signing the entire document, the XML signature standard is feasible to secure complex workflows on a document with multiple signatures. In a proof of concept implementation we demonstrate that verifying and trustworthily displaying of signed documents is realizable in standard Web browsers. The focus of our work are multisigned XML documents that introduce new requirements particularly in the field of presentation. Keywords: Visualisation, WYSIWYS, XML, XML Signature, XPath, XSL Transformation, Web Browser 1 Introduction Electronic data exchange over TCP/IP networks made an overwhelming devel- opment in the past years and resulted in an increased need for security critical services. A major building block to secure documents are digital signatures. They have enormous practical relevance in fundamental services, such as electronic commerce, digital voting or health care. Signing and verifying of electronic documents is very crucial as it preserves the document‘s integrity and authenticity. For this reason, cryptographic algorithms (e.g., RSA, DSA, ECC-DSA) have been studied and are believed to be secure [1]. However, in practice this (single) assumption is insufficient to provide a sophisticated level of security. For instance, in legal proceedings one must also prove that the signed document has been presented correctly, i.e., one must show that the document and its content are clearly assigned to the signer (see e.g. [2]). As the typical user can only verify what she sees3 the content of a digitally signed document must be visualised as well as information about the signer and the verification process (including the certificate chain). 3 This is referred to as the What You See is What you Sign (WYSIWYS) paradigm (see section 2). An additional aggravation is that in today‘s workflows several internal and external business instances process an electronic document. Generally, documents pass a hierarchical network of responsibilities (e.g., employees, supervi- sors) with different roles and access rights. To enhance the workflow (see section 4.1 for a further discussion) new constraints on digital documents have to be made: it is necessary to sign a document manifoldly. Multiple signing means to sign both the entire and certain parts of the document in unspecific order by multiple parties. Due to these requisitions of digital signing several new assumptions of the WYSISWS paradigm have to be made. Solving this paradigm in an user-convenient manner would dramatically enhance the use of digitally signed documents and broaden possible fields of application. Our goal is to demonstrate the benefits of XML Signatures [3] and argue and elaborate that this technology is particularly suitable to design digital processes which require multisigned documents. More precisely, we employ XML and XSLT technologies [4] to fit the above named assumptions of the WYSWYS paradigm and mirror the results in a view, which can be presented, e.g., by standard web browsers. Finally, we show in a proof of concept that the Apache implementation [5] of the XML signature specification can be used for this approach. The remainder sections are structured as follows: in section 2 we briefly introduce the presentation problem and discuss related works in section 3. In section 4 we propose an approach to solve the presentation problem based on XML technologies. In section 5 we prove the feasibility of our idea and provide a proof of concept implementation. Finally, we summarize our work and discuss future work in section 6. 2 The Presentation Problem As mentioned in the introduction, cryptographically proven secure signature algorithms are not sufficient for providing an over-all security in practice; the presentation of the signed document is also decisive. A general problem results from the fact that electronic, respectively, electronically signed documents are processed by machines and displayed in an user-compatible manner, i.e., machines interpret data to the user’s convenience and adequately present it [6]. However, this presentation can be incomplete, incorrect or ambiguous, which yields to a mis-interpretation of significant information including a distorted view of the document‘s integrity and authenticity. This is called the presentation problem [7–9, 6]. For a trustworthily presentation of digital signatures, further assumptions4 have to be fulfilled, which we briefly summarise: The presentation must be definite. In particular, the presentation of signed data must be unambiguous. A verifying person must always be aware of what has been signed. In practice, she is restricted by several factors. Depending on, for example, the displaying device or layout she might be interfered to see the proper document. A frequently mentioned example is the use of white font on 4 We concentrate on basic problems, as a complete discussing would go beyond the paper‘s scope. white background, which hides the actual content. Moreover, active content is (in this context) crucial. On the one hand it provides a level of dynamics, which is required today for a more user-friendly presentation. On the other hand this level of dynamics might inherent the capability to disguise the actual content’s presentation [10]. In addition, the presentation must be transparently. This issue is particularly essential in legal disputes proving non-repudiation: surveyors must trace that the document led to this presentation. In cases of a dispute, the presentation must be reproducible. This is only feasible, if the document format is known, i.e., if each byte and its function is disclosed. For this reason, a definite separation of content and presentation enables a more convenient solution to a higher level of transparency. Perhaps the most important aspect, specifically when considering an entire system, is that the document must be displayed securely. So far, we considered the semantic and syntactic requirements of a document format. We did neither consider the application nor the system, which are threatened by certain attacks (e.g., Trojan horses [7]). An electronic document can only be trustworthily displayed when the intrinsic system is not compromised and sound. In other words, the system and its application are trusted. This is an ultimate prerequisite, or else a digital signature is not tamper-proof. Note that the argumentation is in- vertible: a trusted system, respectively, an application on its own is insufficient to present an electronic document trustworthily. In addition a suitable document format is needed, which enables the document’s content to be displayed unambiguously (as discussed before). In the case of multiple signatures, these requirements become more complex. As a trusted system is not part of this paper, we presuppose an uncompromised and sound system for the present work and concentrate on definitiveness and transparency. 3 Previous Works In the past, several commercial and open document formats/standards have been proposed (see e.g. [11–14]) which are able to present electronically signed documents. A basic approach is to transform the content into an image (see e.g. [15]). The user sees a static effigy of the document. The benefit of this approach is that a static document does not contain any active content, i.e., it circumvents the possibility of altering the content. This approach is from todays point of view impracticable as static documents are hardly editable, hardly processible and unqualified for multi-party business models. Another basic approach is encoded text as used in common signature standards (e.g. S/MIME [12], CMS [13]). However, the encoded text based standards have shortcomings. They define how to embed the signature—even multiple signatures (cf. [14])—into the document format. In contrast to XML, they do not define how to present the content; the presentation depends on its application and, hence, is interpreted product- specifically. The most tangential work was proposed by Scheibelhofer [8]. First of all, he deployed the benefits of XML technologies in the context of signing and verifying electronic documents. He developed an Internet terminal [16] aiming at signing, validating and trustworthily displaying electronic documents. Mainly, his approach is construed for single-signed documents. Our idea goes beyond. Al- though we use similar techniques, we also take into account the usage of multiple XML signatures. Furthermore, we do not provide an architecture for a signing terminal, instead we demonstrate that validating and visualising XML-signed documents is feasible in standard Web browsers. 4 Verification and Visualisation of signed XML documents 4.1 The need for multiple signatures Many business processes require multiple signatures. Usually the responsible persons do not sign at the same time and the same place. For example, the business process of creating an invoice by a fictive software company might require four different signatures: the confirmation of a person that certain goods have been packaged and shipped, the responsible sales person signing for special agreements that have been made with the customer, another signature by a controlling instance (the company’s invoice department) and, finally, the signature of the general manager. Nevertheless, the different process participants are not willing—and partly not able—to sign the document in its completeness due to their restricted areas of responsibility. In this sense the person responsible for packaging for example will not be willing to sign special sales conditions.

Load more